What’s interesting to me is just how quickly companies are doing a 180 on these AI “investments”. You see companies dumping millions into AI projects only to abandon them after like 6 months. You’d think with such large amounts of money on the line there would be some degree of sunk cost fallacy encouraging them to ride things out for longer, but no. When these AI initiatives fail, they fail so spectacularly and so unambiguously that companies pull out fast — faster than corporate inertia usually allows.
The only exceptions seem to be the multi-billion dollar enterprises like Microsoft that can afford to burn money and/or that have shovels to sell to other corps in the brief moment before they crash out.
The reason is simple. It’s become very obvious very quickly that unless you have billions to develop your own model, and a way to lock users in, there is no way you’re ever going to have a chance at making it back. They’re trying to save themselves from making a bad knee jerk reaction.
But they can’t admit that, so they’re trying to hide it however possible and mitigate the financial damage at the cost of anything else they can use.
You can only throw good money after bad money as long as you still have access to good money. I’m sure they’re not abandoning these AI projects because they realize a plan that made no sense six months ago still makes no sense today, but rather because of economic reality - if there’s no money to budget, corporate lack of agility doesn’t matter because the change is enforced externally.
They’re all aware that the chance of success is minimal. However, the one that actually succeeds will achieve an absolutely dominant position both economically and politically. So they’re basically all gambling their companies on being the one that achieves AGI first while being well aware that failure is the most likely outcome. And at this point they can’t pull out because the consequences would be devastating.
Going all in on a tech fundamentally incapable of achieving AGI is a bit dumb, to put it mildly.
We don’t know that yet. While the technology in its current state won’t cut it, there can always be some new breakthrough, that moves the field on.
There is no breakthrough that will make an llm sentient, and dumping the world’s supply of RAM into it 1) doesn’t amount to a “breakthrough”, and 2) will never turn llms into something they’re not.
What you’re saying makes sense when we’re talking about Google or OpenAI or something, but this was a Japanese font company. I really doubt they were chasing AGI — more likely they were using generative AI to make slop fonts and failing miserably.
The only exceptions seem to be the multi-billion dollar enterprises like Microsoft
That was what my answer was referring to.
Gah, my mistake… That’s what I get for replying to messages immediately after waking up…
No sweat. I guess I could have put this more clearly.
“Burn millions and get nothing” Should be put in the flag of generative AI.
Let me ask ChatGPT to generate that…
AI or not this is the trajectory for every private IP holding company. Squeeze as much as possible out of whoever absolutely needs your stuff because you’re not producing anything to service your debts for the acquisitions otherwise.
